21 messages in com.googlegroups.social-graph-apiRe: why is there no sha1 data in outb...
FromSent OnAttachments
ethaniel20 Apr 2008 06:57 
Manish21 Apr 2008 09:53 
ethaniel21 Apr 2008 10:16 
Brad Fitzpatrick21 Apr 2008 10:25 
ethaniel22 Apr 2008 00:33 
Brad Fitzpatrick22 Apr 2008 10:40 
ethaniel22 Apr 2008 23:39 
Brad Fitzpatrick23 Apr 2008 12:02 
Julian Bond23 Apr 2008 12:41 
Brad Fitzpatrick23 Apr 2008 14:24 
Julian Bond24 Apr 2008 00:51 
Danny Ayers24 Apr 2008 02:21 
Julian Bond24 Apr 2008 02:39 
Martin Atkins24 Apr 2008 12:54 
ethaniel25 Apr 2008 13:36 
Brad Fitzpatrick25 Apr 2008 13:44 
ethaniel26 Apr 2008 01:39 
Ben Laurie26 Apr 2008 09:49 
Dale Newfield26 Apr 2008 10:06 
Martin Atkins27 Apr 2008 03:11 
ethaniel30 Apr 2008 11:03 
Subject:Re: why is there no sha1 data in outbound/inbound nodes?
From:Ben Laurie (be@google.com)
Date:04/26/2008 09:49:58 AM
List:com.googlegroups.social-graph-api

On Tue, Apr 22, 2008 at 6:41 PM, Brad Fitzpatrick <brad@google.com> wrote:

I just recently got more resources for the Social Graph API and started moving things around, writing more parsers, improving parsers, getting more data, adding more canonicalization rules, etc.

I switched last night to a new index/version that's been in development for awhile. It's been regularly parsing the most popular documents, but hasn't caught up yet with the long tail. It's parsing now and shouldn't take too much longer. (Sorry, I should've let that finish before flipping the switch to the new version...)

One notable feature of the new version is that it handles the ambiguities between URLs like:

foo.com/blog foo.com/blog/ www.foo.com/blog www.foo.com/blog/

With my protocol hat on I should note that there's no actual requirement for these things to be the same.

... and canonicalizes things down to whichever is the most common.

For instance:

canonicalizes to www.: http://socialgraph.apis.google.com/lookup?q=davidrecordon.com&pretty=1&fme=1

does not: http://socialgraph.apis.google.com/lookup?q=bradfitz.com&pretty=1&fme=1

Now that I'm happy with the parsers and backends and have the resources, I'm going to start working on improving coverage, making sure that for all known URLs, I have a complete picture of the publicly-declared relationships. (I'll be the first to admit that the current coverage isn't perfect.)

- Brad

On Tue, Apr 22, 2008 at 12:34 AM, ethaniel <arka@gmail.com> wrote:

This is strange, I just noticed that my URL above stopped returning any of my Livejournal's nodes. Yesterday it did find my livejournal account by email address. Today it doesn't. Here is what my livejournal looks like:

http://socialgraph.apis.google.com/lookup?q=arkadiynight.livejournal.com&fme=1&edi=1&edo=1&pretty=1&callback=

Anyways, what I was thinking, is - doesn't Livejournal supply sha1sum_mbox data? If it does - then all I could do is enter my email address (sp@ethaniel.com) and it would find my livejournal account through sha1 and show all my livejournal nodes for instance.

Right now it shows only

"nodes_referenced": { "http://k-korzo.livejournal.com/": { "types": [ "friend" ] } },

It would be perfect if it could also show the node's sha1sum_mbox.

What I am trying to achieve is when a person registers on my website, I look his email address in the lookup query, and I would get all his friends's sha1sum_mbox values. Then compare those values with my database and find his possible friends already registered on my website.

Maybe I got this all wrong and this already exists, but not in LiveJournal. And it's the only account I can test on.